Memory Tool — MCP server for Claude Code and Cursor: what it does
Memory Tool — an MCP server that stores and retrieves long-term memories for agents and users across sessions. Covers semantic search and structured JSON filters by user_id, agent_id, date, and more.
no license 107 installs updated 2025-03-28
claude mcp add --transport http mem0-memory-mcp https://mem0-memory-mcp--mem0ai.run.toolsWhat Memory Tool does
Memory Tool connects to Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, and Gemini CLI via a single HTTP endpoint at https://mem0-memory-mcp--mem0ai.run.tools. No local process to run — just register the URL in each client's config.
Nine tools ship with the server. add_memory stores a preference, fact, or conversation snippet; at least one of user_id, agent_id, or run_id is required, or the call fails. search_memories runs a semantic query with filters — narrow by user, agent, or a created_at date range. get_memories pages through records using filters rather than a query string. get_memory fetches a single entry by memory_id, update_memory overwrites its text, and delete_memory removes it after confirming the ID. delete_all_memories clears every record inside a given scope without removing the entity itself; delete_entities drops the entity and cascade-deletes everything under it. list_entities shows which users, agents, apps, or runs currently hold memories.
All writes are scoped to identifiers you supply — the server has no implicit notion of a current user or session.
What it helps with
- persist user preferences across separate agent sessions
- retrieve past facts by meaning rather than exact keyword
- clear all memories for a specific agent or run in one call
- list every active entity that currently holds stored memories
- overwrite a stale fact without touching other records
Example prompts for the agent
- Store that user
aliceprefers dark mode usingadd_memorywithuser_id: alice - Search memories for agent
support-botrelated to proxy configuration - Delete all memories for user
boband remove the entity entirely withdelete_entities
How to install Memory Tool
Platform CLI: remote server
claude mcp add --transport http mem0-memory-mcp https://mem0-memory-mcp--mem0ai.run.toolsPlatform CLI: remote server
codex mcp add memory-tool --url https://mem0-memory-mcp--mem0ai.run.toolsConfig file
{
"mcpServers": {
"memory-tool": {
"url": "https://mem0-memory-mcp--mem0ai.run.tools"
}
}
}Config file
{
"servers": {
"memory-tool": {
"type": "http",
"url": "https://mem0-memory-mcp--mem0ai.run.tools"
}
}
}Platform CLI: remote server
gemini mcp add -t http memory-tool https://mem0-memory-mcp--mem0ai.run.toolsStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Memory Tool
How do I install Memory Tool in Claude Code?
Run claude mcp add --transport http mem0-memory-mcp https://mem0-memory-mcp--mem0ai.run.tools in your terminal. A second alias also works: claude mcp add --transport http memory-tool https://mem0-memory-mcp--mem0ai.run.tools.
What does Memory Tool do?
It gives AI agents a persistent memory store: facts, preferences, and conversation snippets saved across sessions. You can query that store semantically or filter by user_id, agent_id, app_id, run_id, or creation date.
Does Memory Tool require a local server or any installed dependencies?
No local setup is needed. The server runs at https://mem0-memory-mcp--mem0ai.run.tools and accepts HTTP connections directly from supported clients.